On Thu, 31 Oct 2002, Bryan Vyhmeister wrote:

> After spending all that time compiling my kernel, the SE/30 just
> freezes after the "So I sez to him..." quote. I am attaching my kernel
> config and dmesg. Any ideas what could be wrong? I would experiment and
> try more things but it takes so long to compile I figured I would ask
> here first.

> # CPU-related options.
> options 	FPSP
> #options 	FPU_EMULATE

NO! Don't use FPSP on a 68030. FPSP is written in assembler that only
runs on a 68040 (a.k.a. 68040X instructions). At best, it's dead
weight; at worst, any practical floating point arithmetic will
eventually lead an overflow, underflow, or NaN result..., and send
you off into lala land.

> options 	COMPAT_43	# and 4.3BSD

Dead weight.

> options 	COMPAT_M68K4K	# compatibility with NetBSD/m68k4k binaries

Really dead weight.

You could also save a lot of compile time, and some memory, by cutting
out the file systems you don't use.

> # XXX - use only one of ncrscsi or sbc
> #ncrscsi0 at obio?			# SCSI NCR 5380
> sbc0	at obio? flags 0x1		# MI SCSI NCR 5380
> esp0	at obio?			# SCSI NCR 53C9x
> esp1	at obio?			# SCSI NCR 53C9x

Isn't esp only on quadra's?

...but the chief benefits, I should think, would come from eliminating
FPSP, and some of the file systems.

As for the freezing, maybe you're running out of memory, so a smaller
kernel is all you need. You could start with this slightly smaller
than SMALLRAM kernel I built from 1.6 sources for a poster with a
similiar problem:

   ftp://ftp.netbsd.org/pub/NetBSD/arch/mac68k/test/SMALLRAM/SMALLRAM
   ftp://ftp.netbsd.org/pub/NetBSD/arch/mac68k/test/SMALLRAM/netbsd.gz

The SMALLRAM config is optimized for 68030. I had to leave out the
phy's, because it doesn't build otherwise. We should probably build
this variation (and maybe another for 68020's and another for
Quadra's) as part of the release.

Frederick